Estimation is a general problem solving task that has cognitive consequences for students. Many physical and biological phenomena are best modeled using logarithmic scales, but there has been little research that addresses students' understanding of logarithmic scales. Forty junior high school students were asked to solve two types of estimation problems, linear and logarithmic. A microcomputer simulation was used to provide data on students' performance on the two estimation tasks. Past research has shown that this simulation can provide an interesting and reliable method for examining students' learning on an estimation task. Results from this study showed that performance on the linear scale could be described by a general learning curve with improved performance over time. In addition the logarithmic scale results indicated a more complex interaction involving both the number of estimation trials and the position of the estimation “target” on the logarithmic scale. Students' strategy use did not relate to performance on the linear task but did influence performance on the logarithmic task. The results are discussed in terms of students' knowledge of logarithmic scales and their general strategies for problem solving. Implications for teaching and using logarithmic scales in science classes are also discussed.